Skip to main content

Full text of "Regarding x.25"

See other formats


Draft DoD Position 
Regarding X.25 
1. Introduction 
The International Telecommunications Union (ITU) through its 
International Consultative Committee on Telegraphy and Telephony (CCITT) 
has been developing a recommendation for the interfacing of subscriber 
computers to public Packet Switched data networks. This recommendation 
is designated X.25 and specifies the procedures and formats by which 
subscriber data termination equipment (DTE) can exchange packets the 
public data network data circuit termination equipment (DEE). 
Recommendation X.25 also makes reference to lower level-line control 
procedures and electrical interfacing options compatible with the 
so-called X.25 "packet level" interface protocol. The lower level data 
link control procedures include the CCITT/ISO High Level Data Link 
Control procedure (HDLC) and a version of the IBM Bisynchronous Link 
Control procedure ("Bisync"). The electrical inter. face recommendations 
include CCITT recommendations X.21, V.24 and V.35. 
These recommendations, taken together, constitute the body of the X.25 
electrical, link and packet layer protocols which form the lowest three 
levels of the International Standards Organization's Open Systems 
Architecture model. 
In addition to the X.25 recommendation, CCITT has also proposed other 
recommendations for interconnecting public data networks (X.75) and for 
interfacing computer terminals to public data networks (X.28, X.29). 
The principal mode of operation of the X.25 interface is "virtual 
circuit" oriented. The subscriber DTE initiates a virtual circuit 
set-up procedure within the public data net by sending to the public 
data net a connection request packet. Once the virtual circuit is 
established, the public data net returns a connection accept packet and 
the source and destination DTE's can then exchange data packets. 
A receiving DTE, upon receiving a connection request packet can accept 
or reject it. If the connection is rejected, the source DTE is informed 
of this by the public data net. 
The X.25 interface procedures include support for flow control between 
------------------------------<page break>-----------------------------
networks could not be used effectively, if at all, if the only interface 
to them had to adhere to the X.25 recommendations. 
Examples of broadcast or semi-broadcast, DoD packet networks include 
packet satellite, packet radio, Ethernet, Mitrebus, PLRS (position 
location reporting system) and JTIDS (joint tactical information 
distribution system). What is common'about these nets is the fact that 
access to the common communication resource is shared in time, often by 
means of contention detection and resolution methods. This method of 
sharing access to a common communication capacity (e.g. radio channel, 
satellite transponder, coaxial cable) is extremely efficient for large 
numbers of bursty traffic sources, particularly mobile ones. By 
comparison, circuit-like sharing of these resources would be very 
wasteful of the capacity. 
More to the point, however, is the fact that sequencing and integrity 
are services which are not desirable to build into such multi-access 
nets. In order to sequence and maintain the integrity of the packets 
emitted by a source DTE, the subnet must be prepared to retransmit 
packets internally and to buffer them at least at the destination DCE to 
assure re-ordering, if necessary, before delivery to the destination 
DTE. 
In mobile networks, or in systems where local jamming or other hostile 
action is likely, such services merely introduce congestion if packets 
which have arrived at the destination cannot be delivered because the 
"next" one hasn't arrived yet and also introduce large variation in 
packet inter-arrival time at the destination DTE. The attempt to 
maintain integrity may also congest the net'other than at the 
destination if the destination is out of contact (e.g. jammed, 
destroyed, beyond line-of-site, etc.) but this fact isn't yet known to 
the rest of the network. 
The consequences of sequencing within a network are grave if the net is 
intended for real-time data services such as fire control, tracking and 
so on. Furthermore, it has been found that the delay variations 
introduced by attempting to maintain sequencing and integrity within 
such nets makes it impossible to support integrated packet voice 
services as part of the data net. Insisting on only X.25 virtual 
circuit services would render wasted an enormous DoD investment in 
securable, low data-rate packet speech technology. 
Moreover, the point-to-point nature of X.25 renders useless the 
broadcast (multipoint) nature of nets such as Ethernet or the DARPA 
------------------------------<page break>-----------------------------
Atlantic Packet Satellite Net (SATNET). These systems allow a single 
transmission to be received by multiple receivers. This is also a 
feature common to JTIDS, PLRS and a multitude of U.S. Navy 
command/control communication systems. If integrity and sequencing were 
part of the subnet service, the source DCE would have to retransmit 
packets until it had gotten acknowledgements from all destination DCEs 
(or even DTEs). But in hostile condi'tions, not all DTEs or DCEs will be 
accessible. 
Apart from its potentially disastrous imposition of sequencing on 
multi-point communication services, recommendation X.25 also does not 
deal explicitly with communication security and precedence, both of 
which are important to the U.S. Department of Defense. 
As a cosequence of many of these considerations, the majority of the 
experimental U.S. Defense packet networks have been organized around the 
concept of datagrams. A datagram is a finite string of bits containing a 
header which typically indicates the destination address to which the 
bitstring is to be sent, often indicating the source and other relevant 
information such as length, type of transmission service desired, 
precedence and so on. Datagrams typically are transported independently 
of each other by the packet networks, imposing few network mechanisms to 
implement or use the simple and not necessarily reliable or sequential 
datagram service. 
The Department of Defense would not be well-served, if all its packet 
communication systems had to provide service through interfaces meeting 
recommendation X.25 provisions. 
By the same token, however, neither would the Department of Defense be 
well-served if it could not make use of the restricted services provided 
by nets offering only X.25 interfaces. 
During its exploration and development of packet switching techniques, 
the Department of Defense has pursued the development and test of a 
layered protocol architecture which permits a broad range of different 
packet networks types to be interconnected and used end-to-end, despite 
the very significant variations in the classes of services they offered, 
their different interfaces, speed of operation, throughput and maximum 
packet sizes. 
At the heart of this layered architecture is an Internet Protocol (IP) 
which relies only on the most primitive datagram service from each 
constituent network of the Internet System. Figure 2 illustrates the 
------------------------------<page break>-----------------------------
relationships among the protocol layers of the DoO Internet 
Architecture. 
The important differences between the DoD Internet Architecture and the 
networking models developed by CCITT and ISO are: 
D1. The specific existence of an-internet layer. 
D2. The concept of gateways external to the communication subnet. 
D3. The use of encapsulation to transport internet packets through . 
intermediate networks. 
D4. The concept of gateway fragmentation and host (DTE) reassembly. 
D5. The assumption that the basic network service is datagram and 
not virtual circuit. 
DO. The provision for many network interfaces including X.25. 
D7. The explicit provision for security and precedence in the 
internet protocol. 
D8. The coalescing of the OSI Session and Transport layers into a 
.single transport layer and the re-naming of the presentation layer to 
be the utility layer. 
The encapsulation concept, along with the explicit provision of an 
internet protocol layer based on datagram services has made it possible 
to implement, exercise and use daily the protocol hierarchy illustrated 
in Fig 3. 
While only the Internet Protocol, Internet Control Message Protocol and 
Transmission Control Protocol are ratified DoD standards, the other 
protocols are in widespread use in the experimental DoD Internet System 
which includes public networks (U.S. Telenet, UK PSS and IPSS) as well 
as experimental defense networks in the UK (Royal Signals and Radar 
Pilot Packet Switched Network) and Norway (Norwegian Defense Research 
Establishment--NDRENET). 
Creation of the separate gateway system has made it possible to 
incorporate into the DoD Architecture mechanisms for recovering from 
partitioning of a communication subnet by routing traffic through the 
internet gateway system as illustrated in Fig. 4. 
------------------------------<page break>-----------------------------
The encapsulation and fragmentation mech&nisms at the internet level 
have &llowed internet routing to be decoupled from the problem of 
accommodating varying maximum packet sizes in each network. The 
strategy allows the network packet size to be optimized to the 
particular switching and transmission technology (as well as local 
communication and propagation conditions) in each network (Fig. 5). 
By assuming only datagram services from each constituent network, the 
DoD Internet Architecture is able to support a' broader range of 
applications including real-time packet voice. Each internet packet can 
include an indication of the type of service it needs, which might 
trigger the use of virtual circuits on some networks which provide the 
service, but in general, virtual circuit-like service is provided by the 
Transmission Control Protocol at the transport layer, outside the 
collection.of interconnected packet subnets. 
The simplifying datagram service assumption also makes it easier to use 
multiple gateways to share traffic loads since it isn't necessary to 
maintain sequencing and integrity of any particular virtual circuit 
passing across a given network. Packets can be switched to alternate 
gateways as appropriate to share their capacity. This also helps to 
speed up recovery when gateways fail without necessarily requiring 
action on the part of the source DTE (host). 
The details of the security architecture for the DoD Internet System are 
classified, but it can be mentioned that provision for both link-by-link 
and end-to-end security has been made, as well as accommodation for 
multi-level security protection. 
The U.S. Defense Advanced Research Projects Agency (DARPA) is currently 
conducting experiments with its counterparts in Norway (Norwegian 
Defense Research Establishment), the United Kingdom (Royal Signals and 
Radar Establishment), and Germany (DFVLR--the German Air and Space 
Research Agency) on the use and further development of the Internet 
Architecture. These experiments are relevant to the on-going discussion 
among NATO countries concerning NATO standards for packet communication, 
and their results should be factored into any decisions and agreements 
reached within the NATO normalization and standardization process. 
2. The U.S. Department of Defense Position on Recommendation X.25 
In view of the foregoing, it is the U.S. Department of Defense position 
that an acceptable U.S. or NATO standard network architecture must be 
able to make use of, but not be limited to, networks providing 
------------------------------<page break>-----------------------------
interfaces meeting the CCITT Recommendation X.25. In particular, 
provision for non-virtual circuit modes of operation are considered 
mandatory to support transaction or real-time applications in an 
internetwork environment. 
------------------------------<page break>-----------------------------
FIGURE 1 ILLUSTRATES THE APPROXIMATE 
RELATIONSHIPS AMONG THE VARIOUS RECOMMENDATIONS 
DTE/HOST 
I X.29 ! =X.121 ADDRESSING RECOMMENDATION 
X.25 PACKET LEVEL 
HDLC/BISYNC TERMINAL 
I STE I 
X,21... 
TERMINAL I/F 
DTE/PAD 
-- X,75 GATEWAY PROCEDURES 
STE 
DTE/PAD 
PUBLIC 
DATA 
NET 
DCE 
X.21.,. 
TERMINAL 
CCI'I-r PROTOCOL RECOMMENDATIONS 
FIGURE I 
------------------------------<page break>-----------------------------
------------------------------<page break>-----------------------------
! 
! 
------------------------------<page break>-----------------------------